No abstract

No abstract

With the development of location-based services, such as the Global Positioning System and Radio Frequency Identification, a great deal of trajectory data can be collected. Therefore, how to mine knowledge from these data has become an attractive topic. In this paper, we propose an efficient trajectory-clustering algorithm based on an index tree. Firstly, an index tree is proposed to store trajectories and their similarity matrix, with which trajectories can be retrieved efficiently; secondly, a new conception of trajectory structure is introduced to analyse both the internal and external features of trajectories; then, trajectories are partitioned into trajectory segments according to their corners; furthermore, the similarity between every trajectory segment pairs is compared by presenting the structural similarity function; finally, trajectory segments are grouped into different clusters according to their location in the different levels of the index tree. Experimental results on real data sets demonstrate not only the efficiency and effectiveness of our algorithm, but also the great flexibility that feature sensitivity can be adjusted by different parameters, and the cluster results are more practically significant.

Neutrophil lymphocyte ratio (NLR) and platelet lymphocyte ratio (PLR) had been analysed in many kind of tumours, but its role of predict the oesophageal squamous cell carcinoma (ESCC) patients' prognosis was not reach a consensus. Relationship between NLR, PLR and ESCC located in the middle or lower segment was evaluated. 317 patients with ESCC who underwent attempted curative oesophagectomy were analysed in this study. 157 and 98 patients had elevated NLR and PLR respectively (NLR >3.3 and PLR >150). The median overall survival time (OS) and disease-free survival (DFS) was 34.1 and 19.2 months respectively. Multivariate analysis found PLR >150 (P = 0.018, HR 1.426, 95%CI 1.063-1.912) accompanied by male, lymphatic metastases, tumour size more than 3 cm, tumour located at middle segment and poor differentiation were associated with significantly worse DFS. Meanwhile, gender, lymphatic metastases, tumour location and differentiation along with PLR >150 (P = 0.003, HR 1.595, 95% CI 1.172-2.170) and NLR>3.3 (P = 0.039, HR 1.367, 95% CI 1.015-1.840) were all independent prognostic factors for OS. Preoperative NLR and PLR might be used as predictive factors in patients with ESCC. For DFS, elevated PLR compared to NLR may have an advantage to indicate poor prognosis.